ISC · Information Systems and Data Management · 9 practice questions
Single Points of Failure: Match the Infrastructure Control
An availability control must protect the component whose failure would stop processing. Below, sort infrastructure changes by whether they address the outage cause or leave a single point of failure.
Try one first
Hint
Focus on the stated objective: reducing downtime caused by failure of infrastructure components already identified in the stem.
Answer C. The environment currently has single points of failure: one server and one switch. Adding redundant servers and network equipment with failover capability directly improves availability by allowing processing to continue if one component fails. Because the stated objective is to reduce unplanned downtime from hardware or network component failure, redundancy is the best response.
Why not A: Backups protect against data loss and aid recovery, but they do not keep the system running during a server or switch failure. Backups address restore and recovery, not immediate availability.
Why not B: This is tempting because capacity upgrades can improve performance, but increasing resources on a single server does not remove the single point of failure; the server could still fail and cause downtime.
Why not D: Multifactor authentication improves access security but does nothing to prevent or mitigate hardware or network component failures that cause downtime.
Sort it
The control supplies alternate processing, connectivity, or power for the stated failure.
The control improves recovery, capacity, or a different component instead of protecting against the stated failure.
Apparently redundant resources still depend on one component or path that can disable them together.
| Item | Goes to |
|---|---|
| Server hardware failure: North adds a second server with synchronized production data and automatic failover. | Targets the failureThe second server can take over processing when the original server fails. |
| North runs two application VMs on one physical host and calls the arrangement highly available. | Shared failure point remainsBoth VMs stop if their shared physical host fails. Separate VMs are not independent hardware. |
| Server hardware failure: North increases full backups from nightly to hourly. | Does not target the failureFresher backups reduce potential data loss but do not provide a running replacement server. |
| Brief power dips: North installs a UPS for critical servers and network devices. | Targets the failureThe UPS supplies short-term power so brief interruptions do not force those devices to restart. |
| North clusters servers on separate hosts, but both hosts require one shared storage array with no redundancy. | Shared failure point remainsFailure of the shared storage can stop processing on both hosts despite successful compute failover. |
| Core switch failure: North adds a second core switch, independent network paths, and automatic failover. | Targets the failureTraffic can use the alternate switch and path instead of waiting for replacement hardware. |
| Whole-server hardware failure: North installs RAID disk mirroring inside the existing server. | Does not target the failureDisk mirroring protects against a drive failure, not failure of the entire server. |
| WAN carrier outage: North adds a connection from another carrier over an independent route with automatic failover. | Targets the failureThe alternate connection preserves connectivity when the original carrier becomes unavailable. |
| North uses two carriers, but both connections travel over the same fiber route. | Shared failure point remainsDamage to that route can disable both connections. Different carrier names do not establish path independence. |
| Stateless application server failure: North places interchangeable servers behind a load balancer with health checks. | Targets the failureHealth checks remove the failed server from service while the remaining server handles requests. |
| Server hardware failure: North upgrades the existing server's CPU and memory. | Does not target the failureThe upgrade increases capacity but leaves processing dependent on the same physical server. |
| North has multiple application servers, but all user requests must pass through one load balancer with no failover. | Shared failure point remainsThe application servers are redundant, but failure of their sole load balancer can still block access. |
Key points
- Remember: Ask what can still stop processing, not how many servers the company owns.
- Health checks let a load balancer stop sending requests to a failed application server.
- Backups support restoration; they do not supply live processing capacity.
- Two carriers are not independent if both connections depend on one fiber route.
How the exam traps you
- Choose more frequent backups when the goal is continued operation. Choose failover capacity for the production component that could fail.
- Choose a faster server when the concern is server failure. More capacity leaves the same physical server as a single point of failure.
- Treat redundancy in one tier as protection for the whole system. Check shared hosts, storage, load balancers, and network paths.
Question 2
Hint
Consider what counts as a voting member in a majority‑quorum cluster, does data replication itself provide a vote?
Answer B. Majority‑quorum clusters need a majority of voting members to be available; with only two voting nodes a single-node failure produces no majority. An external quorum witness supplies the deciding third vote, allowing a supported two-node configuration to maintain quorum. Storage replication protects data and enables VM failover but does not act as a voting member, so replication alone does not resolve quorum/tie‑breaking risk.
Why not A: Tempting because replication and automatic failover can preserve VM data and restart workloads on surviving hardware; however, this ignores cluster membership rules. Replication does not provide a quorum vote, so the cluster can lose majority (or suspend services) even if data is intact.
Why not C: Tempting because rack separation and synchronous offsite replication reduce correlated failures and protect data availability. However, physical separation and replication do not create a third voting member, so they do not resolve the underlying quorum/tie‑breaking problem.
Why not D: Tempting because keeping three nodes is the simplest way to preserve quorum and avoid additional components. But this answer is overly absolute: it ignores the vendor-supported mitigation (an external quorum witness) that can safely allow a two-node deployment while addressing the quorum risk.
Question 3
Hint
Choose the option that eliminates a single server as the single point of failure and enables near-immediate continuity without full site failover.
Answer D. Using a load balancer with multiple identical stateless servers removes the single application server as a single point of failure; health checks allow the balancer to stop sending traffic to a failed server so requests continue to be served with minimal interruption. This directly addresses a single-server failure within the same data center without requiring full site failover.
Why not A: Cross-site synchronous replication and automatic failover provide site-level redundancy, but they are more complex and typically intended for site outages or disaster recovery. This is overkill for a single-server failure within the same data center and may still involve noticeable interruption during failover.
Why not B: RAID 1 protects against a disk failure, but it does not prevent outages from CPU failure, OS crashes, or an application process failure on the server; it therefore doesn't meet the availability requirement by itself.
Why not C: Backups protect against data loss and support recovery, but they address longer RTO/RPO goals and won't provide near-immediate continuity for user requests after a server failure.
Question 4
Hint
Focus on what disk mirroring duplicates and what it does not.
Answer A. RAID 1 mirrors data across drives, so it helps if one physical disk fails. However, it is not a substitute for backups because a file deletion, ransomware encryption, or data corruption can be copied to both mirrored drives; RAID improves hardware fault tolerance, not full data recovery capability.
Why not B: This choice confuses RAID behavior. RAID 1 is specifically designed so the system can usually keep running after one drive fails by using the surviving mirrored drive, so saying it prevents continued operation is the opposite of its basic benefit.
Why not C: This is an overgeneralization: having two copies on the same server is not the same as an offsite backup. Both copies can be lost to theft, fire, ransomware, accidental deletion, or corruption, so offsite/backups and RAID address different risks.
Why not D: This distractor confuses storage with networking. RAID 1 is a disk redundancy method focused on hardware fault tolerance; it does not primarily affect network bandwidth and its main value is protection against a disk failure, not network performance.
Question 5
Hint
Which redundancy was removed and does the cloud setup include multi‑region or local redundancy? Also consider the firm's reliance on internet/VPN for access.
Answer D. Consolidating all data and backups into one cloud region and removing local copies creates dependence on that single region and on the network path. Without cross‑region replication or on‑site redundancy, a regional outage or internet/VPN failure can produce prolonged downtime, a classic single point of failure.
Why not A: Tempting because cloud providers offer standardized controls, but compliance depends on configuration, data residency, contracts, and access controls. Centralizing in one region does not automatically reduce regulatory risk and can introduce new exposures.
Why not B: Tempting because cloud platforms advertise high availability, but RTO gains require explicit disaster‑recovery or multi‑region configurations. With no cross‑region replication and no local backups, a region‑level outage or connectivity loss may actually extend recovery time rather than shorten it.
Why not C: Tempting because centralization can reduce on‑site management, but moving access from LAN to internet/VPN typically increases latency and can degrade on‑site performance. Without WAN acceleration or edge caching (not deployed here), centralization alone is unlikely to improve access speed.
Question 6
Hint
Decide whether the primary goal is minimizing downtime with automatic switching (HA) versus recovering after a major event or simply keeping backups.
Answer C. The configuration (synchronous replication across sites plus automatic failover with only brief interruptions) is designed to minimize service downtime and provide continuous operation, which is the goal of high availability. Disaster recovery and backups focus on restoring after larger outages or data loss and may tolerate longer downtime, while fault tolerance implies seamless masking of failures without interruption.
Why not A: Tempting because geographically separated sites and failover are elements of disaster recovery. It's incorrect here because disaster recovery emphasizes restoring operations after major incidents and often accepts longer downtime, whereas the stem emphasizes minimizing downtime via automatic failover (high availability).
Why not B: Tempting since data protection is mentioned, but backups are periodic copies used for data restoration and long-term retention; they do not provide near-real-time synchronous replication and automatic failover that minimize downtime.
Why not D: Tempting because fault tolerance aims to keep services running despite component failures. It's incorrect here because true fault tolerance masks failures without any interruption; the stem describes brief interruptions during automatic failover, which aligns with high availability rather than full fault tolerance.
Question 7
Hint
Focus on which infrastructure control is intended to bridge very short power interruptions.
Answer B. A UPS is designed to provide short-term power during brief outages or voltage dips, which directly addresses the stated problem. It can keep systems and network devices running through short interruptions and may also provide enough time for an orderly shutdown if the utility power is not restored quickly; the other options do not provide immediate power continuity.
Why not A: Backups are useful for recovering data after a failure but do not prevent servers and network devices from restarting during short power dips or keep services available during the interruption. This option addresses recovery, not immediate power continuity.
Why not C: RAID improves resilience against disk failures and can help availability when a drive fails, but it does not supply electrical power or prevent restarts caused by brief utility interruptions. The root cause here is power continuity, not storage redundancy.
Why not D: Multifactor authentication strengthens access security and helps prevent unauthorized access, but it does not address service outages caused by power dips. This choice confuses security controls with availability controls.
Question 8
Hint
Identify which infrastructure elements could be disabled by one local event and thereby disrupt both customer connectivity and the ability to fail over.
Answer B. The root infrastructure deficiency was the shared geographic and network dependencies: both DCs are in the same metro area and both rely on a single regional ISP fiber route. That created single points of failure so one local event (the storm and fiber damage) could and did disrupt customer connectivity for both sites. Asynchronous replication lag and an untested DR plan worsened recovery and data completeness but are secondary to the lack of true redundancy.
Why not A: This is tempting because backups are critical to recovery, but it contradicts the stem: nightly backups were written to tape and stored in a physically separate offsite vault, so unavailable backups were not the primary cause of the outage.
Why not C: Asynchronous replication lag did cause partial data loss and explains why some recent transactions were missing after failover, making it an attractive secondary explanation. However, it does not explain why customers experienced a broad multi-hour connectivity outage when both DCs lost upstream network access due to a shared dependence on the same fiber/ISP.
Why not D: An untested DR plan can delay and complicate recovery and is a plausible contributing factor, but it is secondary here: even a tested plan could be hampered if both DCs and upstream connectivity were simultaneously taken down by the same local event because of insufficient redundancy.
Question 9
Hint
Focus on what happens if the hardware itself stops working, not on user access or transaction entry.
Answer A. Multiple critical systems depend on a single physical server with no secondary hardware or failover. That creates a single point of failure: a hardware fault would render those systems unavailable until replacement hardware is in place and data is restored. Offsite backups aid recovery but do not prevent the immediate availability risk.
Why not B: This distractor confuses application controls with infrastructure issues. The stem gives no evidence of missing input validation or authorization controls; the problem described is lack of redundancy and availability.
Why not C: Mentioning accounting systems can mislead candidates, but there is no information about user roles, permissions, or incompatible duties. The issue is the server-dependent availability of systems, not staff role design.
Why not D: Offsite backups are a recovery control and do not equate to audit logs or an audit trail. The stem does not indicate missing logging; it indicates downtime risk from a single-server architecture.
Common questions
Do backups eliminate a single point of failure?
No. Backups let you restore data, but they do not keep processing running when the only production server fails.
Do two virtual machines provide high availability?
Not if both depend on one physical host. That host can fail and stop both virtual machines.
Does a UPS protect against server hardware failure?
No. A UPS supplies short-term power during interruptions. A second server with failover addresses failure of the server itself.
Practice ISC like the real exam
The free ChatCPA simulator: real exam layout, timed testlets, starting with a question on this topic. No account needed to start.
Open the free simulator →More on IT infrastructure
Questions from the ChatCPA bank of 17,658 CPA exam questions, each with a written reason for every wrong answer. ChatCPA is built by Nicholas Miller, CPA (Oregon #14907). How these pages are made. Spot an error? Tell us.